-#include <stdlib.h>
-#include <stdio.h>
-#include <string.h>
-#include <sys/time.h>
-#include <unistd.h>
-#include <sys/socket.h>
-#include <netdb.h>
-#include <signal.h>
-#include <ctype.h>
-#include <assert.h>
-
-#include <yaz/marcdisp.h>
-#include <yaz/comstack.h>
-#include <yaz/tcpip.h>
-#include <yaz/proto.h>
-#include <yaz/readconf.h>
-#include <yaz/pquery.h>
-#include <yaz/yaz-util.h>
-#include <yaz/nmem.h>
-
-#if HAVE_CONFIG_H
-#include "cconfig.h"
-#endif
-
-#define USE_TIMING 0
-#if USE_TIMING
-#include <yaz/timing.h>
-#endif
-
-#include "pazpar2.h"
-#include "eventl.h"
-#include "http.h"
-#include "termlists.h"
-#include "reclists.h"
-#include "relevance.h"
-#include "config.h"
-
-#define MAX_CHUNK 15
-
-static void client_fatal(struct client *cl);
-static void connection_destroy(struct connection *co);
-static int client_prep_connection(struct client *cl);
-static void ingest_records(struct client *cl, Z_Records *r);
-static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
-void session_alert_watch(struct session *s, int what);
-
-IOCHAN channel_list = 0; // Master list of connections we're handling events to
-
-static struct connection *connection_freelist = 0;
-static struct client *client_freelist = 0;
-
-static struct host *hosts = 0; // The hosts we know about
-static struct database *databases = 0; // The databases we know about
-
-static char *client_states[] = {
- "Client_Connecting",
- "Client_Connected",
- "Client_Idle",
- "Client_Initializing",
- "Client_Searching",
- "Client_Presenting",
- "Client_Error",
- "Client_Failed",
- "Client_Disconnected",
- "Client_Stopped"
-};
-
-// Note: Some things in this structure will eventually move to configuration
-struct parameters global_parameters =
-{
- 0,
- 0,
- 30,
- "81",
- "Index Data PazPar2 (MasterKey)",
- VERSION,
- 600, // 10 minutes
- 60,
- 100,
- MAX_CHUNK,
- 0,
- 0,
- 0,
- 0
-};
-
-static int send_apdu(struct client *c, Z_APDU *a)
-{
- struct connection *co = c->connection;
- char *buf;
- int len, r;
-
- if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
- {
- odr_perror(global_parameters.odr_out, "Encoding APDU");
- abort();
- }
- buf = odr_getbuf(global_parameters.odr_out, &len, 0);
- r = cs_put(co->link, buf, len);
- if (r < 0)
- {
- yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
- return -1;
- }
- else if (r == 1)
- {
- fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
- exit(1);
- }
- odr_reset(global_parameters.odr_out); /* release the APDU structure */
- co->state = Conn_Waiting;
- return 0;
-}
-
-
-static void send_init(IOCHAN i)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
-
- a->u.initRequest->implementationId = global_parameters.implementationId;
- a->u.initRequest->implementationName = global_parameters.implementationName;
- a->u.initRequest->implementationVersion =
- global_parameters.implementationVersion;
- ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
- ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
- ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
-
- ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
- ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
- ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
- if (send_apdu(cl, a) >= 0)
- {
- iochan_setflags(i, EVENT_INPUT);
- cl->state = Client_Initializing;
- }
- else
- cl->state = Client_Error;
- odr_reset(global_parameters.odr_out);
-}
-
-static void send_search(IOCHAN i)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- struct session *se = cl->session;
- struct database *db = cl->database;
- Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
- int ndb, cerror, cpos;
- char **databaselist;
- Z_Query *zquery;
- struct ccl_rpn_node *cn;
- int ssub = 0, lslb = 100000, mspn = 10;
-
- yaz_log(YLOG_DEBUG, "Sending search");
-
- cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
- if (!cn)
- return;
- a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
- sizeof(Z_Query));
- zquery->which = Z_Query_type_1;
- zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
- ccl_rpn_delete(cn);
-
- for (ndb = 0; db->databases[ndb]; ndb++)
- ;
- databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
- for (ndb = 0; db->databases[ndb]; ndb++)
- databaselist[ndb] = db->databases[ndb];
-
- a->u.presentRequest->preferredRecordSyntax =
- yaz_oidval_to_z3950oid(global_parameters.odr_out,
- CLASS_RECSYN, VAL_USMARC);
- a->u.searchRequest->smallSetUpperBound = &ssub;
- a->u.searchRequest->largeSetLowerBound = &lslb;
- a->u.searchRequest->mediumSetPresentNumber = &mspn;
- a->u.searchRequest->resultSetName = "Default";
- a->u.searchRequest->databaseNames = databaselist;
- a->u.searchRequest->num_databaseNames = ndb;
-
- if (send_apdu(cl, a) >= 0)
- {
- iochan_setflags(i, EVENT_INPUT);
- cl->state = Client_Searching;
- cl->requestid = se->requestid;
- }
- else
- cl->state = Client_Error;
-
- odr_reset(global_parameters.odr_out);
-}
-
-static void send_present(IOCHAN i)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
- int toget;
- int start = cl->records + 1;
-
- toget = global_parameters.chunk;
- if (toget > global_parameters.toget - cl->records)
- toget = global_parameters.toget - cl->records;
- if (toget > cl->hits - cl->records)
- toget = cl->hits - cl->records;
-
- yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
-
- a->u.presentRequest->resultSetStartPoint = &start;
- a->u.presentRequest->numberOfRecordsRequested = &toget;
-
- a->u.presentRequest->resultSetId = "Default";
-
- a->u.presentRequest->preferredRecordSyntax =
- yaz_oidval_to_z3950oid(global_parameters.odr_out,
- CLASS_RECSYN, VAL_USMARC);
-
- if (send_apdu(cl, a) >= 0)
- {
- iochan_setflags(i, EVENT_INPUT);
- cl->state = Client_Presenting;
- }
- else
- cl->state = Client_Error;
- odr_reset(global_parameters.odr_out);
-}
-
-static void do_initResponse(IOCHAN i, Z_APDU *a)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- Z_InitResponse *r = a->u.initResponse;
-
- yaz_log(YLOG_DEBUG, "Received init response");
-
- if (*r->result)
- {
- cl->state = Client_Idle;
- }
- else
- cl->state = Client_Failed; // FIXME need to do something to the connection
-}
-
-static void do_searchResponse(IOCHAN i, Z_APDU *a)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- struct session *se = cl->session;
- Z_SearchResponse *r = a->u.searchResponse;
-
- yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
-
- if (*r->searchStatus)
- {
- cl->hits = *r->resultCount;
- se->total_hits += cl->hits;
- if (r->presentStatus && !*r->presentStatus && r->records)
- {
- yaz_log(YLOG_DEBUG, "Records in search response");
- cl->records += *r->numberOfRecordsReturned;
- ingest_records(cl, r->records);
- }
- cl->state = Client_Idle;
- }
- else
- { /*"FAILED"*/
- cl->hits = 0;
- cl->state = Client_Error;
- if (r->records) {
- Z_Records *recs = r->records;
- if (recs->which == Z_Records_NSD)
- {
- yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
- cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
- cl->state = Client_Error;
- }
- }
- }
-}
-
-char *normalize_mergekey(char *buf)
-{
- char *p = buf, *pout = buf;
-
- while (*p)
- {
- while (*p && !isalnum(*p))
- p++;
- while (isalnum(*p))
- *(pout++) = tolower(*(p++));
- if (*p)
- *(pout++) = ' ';
- while (*p && !isalnum(*p))
- p++;
- }
- if (buf != pout)
- *pout = '\0';
-
- return buf;
-}
-
-
-#ifdef GAGA
-// FIXME needs to be generalized. Should flexibly generate X lists per search
-static void extract_subject(struct session *s, const char *rec)
-{
- const char *field, *subfield;
-
- while ((field = find_field(rec, "650")))
- {
- rec = field;
- if ((subfield = find_subfield(field, 'a')))
- {
- char *e, *ef;
- char buf[1024];
- int len;
-
- ef = index(subfield, '\n');
- if (!ef)
- return;
- if ((e = index(subfield, '\t')) && e < ef)
- ef = e;
- while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
- ef--;
- len = ef - subfield;
- assert(len < 1023);
- memcpy(buf, subfield, len);
- buf[len] = '\0';
-#ifdef FIXME
- if (*buf)
- termlist_insert(s->termlist, buf);
-#endif
- }
- }
-}
-#endif
-
-static void add_facet(struct session *s, const char *type, const char *value)
-{
- int i;
-
- for (i = 0; i < s->num_termlists; i++)
- if (!strcmp(s->termlists[i].name, type))
- break;
- if (i == s->num_termlists)
- {
- if (i == SESSION_MAX_TERMLISTS)
- {
- yaz_log(YLOG_FATAL, "Too many termlists");
- exit(1);
- }
- s->termlists[i].name = nmem_strdup(s->nmem, type);
- s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
- s->num_termlists = i + 1;
- }
- termlist_insert(s->termlists[i].termlist, value);
-}
-
-int yaz_marc_write_xml();
-
-static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
-{
- struct conf_retrievalprofile *rprofile = cl->database->rprofile;
- struct conf_retrievalmap *m;
- xmlNode *res;
- xmlDoc *rdoc;
-
- // First normalize to XML
- if (rprofile->native_syntax == Nativesyn_iso2709)
- {
- char *buf;
- int len;
- if (rec->which != Z_External_octet)
- {
- yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
- return 0;
- }
- buf = (char*) rec->u.octet_aligned->buf;
- len = rec->u.octet_aligned->len;
- if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
- {
- yaz_log(YLOG_WARN, "Failed to decode MARC");
- return 0;
- }
- if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
- "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
- {
- yaz_log(YLOG_WARN, "Failed to encode as XML");
- return 0;
- }
- rdoc = xmlNewDoc("1.0");
- xmlDocSetRootElement(rdoc, res);
- }
- else
- {
- yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
- exit(1);
- }
- for (m = rprofile->maplist; m; m = m->next)
- {
- xmlDoc *new;
- if (m->type != Map_xslt)
- {
- yaz_log(YLOG_WARN, "Unknown map type");
- return 0;
- }
- if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
- {
- yaz_log(YLOG_WARN, "XSLT transformation failed");
- return 0;
- }
- xmlFreeDoc(rdoc);
- rdoc = new;
- }
- if (global_parameters.dump_records)
- {
- fprintf(stderr, "Record:\n----------------\n");
- xmlDocFormatDump(stderr, rdoc, 1);
- }
- return rdoc;
-}
-
-static struct record *ingest_record(struct client *cl, Z_External *rec)
-{
- xmlDoc *xdoc = normalize_record(cl, rec);
- xmlNode *root, *n;
- struct record *res;
- struct record_cluster *cluster;
- struct session *se = cl->session;
- xmlChar *mergekey, *mergekey_norm;
- xmlChar *type;
- xmlChar *value;
- struct conf_service *service = global_parameters.server->service;
-
- if (!xdoc)
- return 0;
-
- root = xmlDocGetRootElement(xdoc);
- if (!(mergekey = xmlGetProp(root, "mergekey")))
- {
- yaz_log(YLOG_WARN, "No mergekey found in record");
- xmlFreeDoc(xdoc);
- return 0;
- }
-
- res = nmem_malloc(se->nmem, sizeof(struct record));
- res->next = 0;
- res->target_offset = -1;
- res->term_frequency_vec = 0;
- res->metadata = nmem_malloc(se->nmem,
- sizeof(struct record_metadata*) * service->num_metadata);
- bzero(res->metadata, sizeof(struct record_metadata*) * service->num_metadata);
- res->relevance = 0;
-
- mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
- xmlFree(mergekey);
- normalize_mergekey(mergekey_norm);
-
- cluster = reclist_insert(se->reclist, res, mergekey_norm);
- if (!cluster)
- {
- /* no room for record */
- xmlFreeDoc(xdoc);
- return 0;
- }
- relevance_newrec(se->relevance, cluster);
-
- type = value = 0;
- for (n = root->children; n; n = n->next)
- {
- if (type)
- xmlFree(type);
- if (value)
- xmlFree(value);
- type = value = 0;
-
- if (n->type != XML_ELEMENT_NODE)
- continue;
- if (!strcmp(n->name, "metadata"))
- {
- type = xmlGetProp(n, "type");
- value = xmlNodeListGetString(xdoc, n->children, 0);
- struct conf_metadata *md = 0;
- struct record_metadata **wheretoput, *newm;
- int imeta;
-
- // First, find out what field we're looking at
- for (imeta = 0; imeta < service->num_metadata; imeta++)
- if (!strcmp(type, service->metadata[imeta].name))
- {
- md = &service->metadata[imeta];
- break;
- }
- if (!md)
- {
- yaz_log(YLOG_WARN, "Ignoring unknown metadata element: %s", type);
- continue;
- }
-
- // Find out where we are putting it
- if (md->merge == Metadata_merge_no)
- wheretoput = &res->metadata[imeta];
- else
- wheretoput = &cluster->metadata[imeta];
-
- // Put it there
- newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
- newm->next = 0;
- if (md->type == Metadata_type_generic)
- {
- newm->data.text = nmem_strdup(se->nmem, value);
- }
- else
- {
- yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type);
- continue;
- }
- if (md->merge == Metadata_merge_unique)
- {
- struct record_metadata *mnode;
- for (mnode = *wheretoput; mnode; mnode = mnode->next)
- if (!strcmp(mnode->data.text, mnode->data.text))
- break;
- if (!mnode)
- {
- newm->next = *wheretoput;
- *wheretoput = newm;
- }
- }
- else if (md->merge == Metadata_merge_longest)
- {
- if (!*wheretoput ||
- strlen(newm->data.text) > strlen((*wheretoput)->data.text))
- *wheretoput = newm;
- }
- else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no)
- {
- newm->next = *wheretoput;
- *wheretoput = newm;
- }
- else
- yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
-
- if (md->rank)
- relevance_countwords(se->relevance, cluster, value, md->rank);
- if (md->termlist)
- add_facet(se, type, value);
- xmlFree(type);
- xmlFree(value);
- type = value = 0;
- }
- else
- yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
- }
-
- xmlFreeDoc(xdoc);
-
- relevance_donerecord(se->relevance, cluster);
- se->total_records++;
-
- return res;
-}
-
-static void ingest_records(struct client *cl, Z_Records *r)
-{
-#if USE_TIMING
- yaz_timing_t t = yaz_timing_create();
-#endif
- struct record *rec;
- struct session *s = cl->session;
- Z_NamePlusRecordList *rlist;
- int i;
-
- if (r->which != Z_Records_DBOSD)
- return;
- rlist = r->u.databaseOrSurDiagnostics;
- for (i = 0; i < rlist->num_records; i++)
- {
- Z_NamePlusRecord *npr = rlist->records[i];
-
- if (npr->which != Z_NamePlusRecord_databaseRecord)
- {
- yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
- continue;
- }
-
- rec = ingest_record(cl, npr->u.databaseRecord);
- if (!rec)
- continue;
- }
- if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
- session_alert_watch(s, SESSION_WATCH_RECORDS);
-
-#if USE_TIMING
- yaz_timing_stop(t);
- yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
- yaz_timing_get_real(t), yaz_timing_get_user(t),
- yaz_timing_get_sys(t));
- yaz_timing_destroy(&t);
-#endif
-}
-
-static void do_presentResponse(IOCHAN i, Z_APDU *a)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- Z_PresentResponse *r = a->u.presentResponse;
-
- if (r->records) {
- Z_Records *recs = r->records;
- if (recs->which == Z_Records_NSD)
- {
- yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
- cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
- cl->state = Client_Error;
- }
- }
-
- if (!*r->presentStatus && cl->state != Client_Error)
- {
- yaz_log(YLOG_DEBUG, "Good Present response");
- cl->records += *r->numberOfRecordsReturned;
- ingest_records(cl, r->records);
- cl->state = Client_Idle;
- }
- else if (*r->presentStatus)
- {
- yaz_log(YLOG_WARN, "Bad Present response");
- cl->state = Client_Error;
- }
-}
-
-static void handler(IOCHAN i, int event)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- struct session *se = 0;
-
- if (cl)
- se = cl->session;
- else
- {
- yaz_log(YLOG_WARN, "Destroying orphan connection");
- connection_destroy(co);
- return;
- }
-
- if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
- {
- int errcode;
- socklen_t errlen = sizeof(errcode);
-
- if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
- &errlen) < 0 || errcode != 0)
- {
- client_fatal(cl);
- return;
- }
- else
- {
- yaz_log(YLOG_DEBUG, "Connect OK");
- co->state = Conn_Open;
- if (cl)
- cl->state = Client_Connected;
- }
- }
-
- else if (event & EVENT_INPUT)
- {
- int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
-
- if (len < 0)
- {
- yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
- connection_destroy(co);
- return;
- }
- else if (len == 0)
- {
- yaz_log(YLOG_WARN, "EOF reading from Z server");
- connection_destroy(co);
- return;
- }
- else if (len > 1) // We discard input if we have no connection
- {
- co->state = Conn_Open;
-
- if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
- {
- Z_APDU *a;
-
- odr_reset(global_parameters.odr_in);
- odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
- if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
- {
- client_fatal(cl);
- return;
- }
- switch (a->which)
- {
- case Z_APDU_initResponse:
- do_initResponse(i, a);
- break;
- case Z_APDU_searchResponse:
- do_searchResponse(i, a);
- break;
- case Z_APDU_presentResponse:
- do_presentResponse(i, a);
- break;
- default:
- yaz_log(YLOG_WARN, "Unexpected result from server");
- client_fatal(cl);
- return;
- }
- // We aren't expecting staggered output from target
- // if (cs_more(t->link))
- // iochan_setevent(i, EVENT_INPUT);
- }
- else // we throw away response and go to idle mode
- {
- yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
- cl->state = Client_Idle;
- }
- }
- /* if len==1 we do nothing but wait for more input */
- }
-
- if (cl->state == Client_Connected) {
- send_init(i);
- }